"We get asked about AI at every board. I do not have an answer I would want to defend."
What was approved, what was built, what is in daily use, and what people started using without asking.
For each case: what it has delivered against where you started, including any revenue; what it cost to build, buy or partner, and what it costs to run at your real volumes; and what it does to your own capability, your data and the lock-in you've accepted.
Continue, stop, restart on different assumptions, or wait with a set point to look again. Improving one of the three usually costs you on another, so each call sets out the balance and the assumptions behind it.
I draft with you a short written position for the board: what you build, what you buy and what you keep in-house. There's a shareholder version too, if you need one.
This looks back at the AI you already run. To find and test new use cases, see the AI Use Case Lab.
Not included: building pipelines, technical audit of the models, tool selection and acceptable-use policy.
Fixed price, agreed after a short call. How I work →